Notes on line length revisited: Does recent research debunk the idea of an ideal line length?

A few months ago I wrote about my approach to web typography basics. I’ve been plugging away at this website for more than 13 years, and I’ve been interested in screen typography for even longer. You couldn’t describe any expertise I’ve amassed over this time as academic – as with so much digital knowledge, it’s been gained informally through blogs, social media and the odd bit of research – but I feel I have some useful knowledge. I know how to design a readable blog.

You can therefore imagine my excitement when I came across an article by Mary Dyson questioning some of the assumptions – which I’ve held – about line length. It summarises the most recent substantial research, a PhD by Adam Parker of Bournemouth University entitled The return-sweep in reading (download the 1MB PhD PDF), comparing it to the established wisdom of an influential journal article from 1940 Influence of line width on eye movements by DG Paterson and MA Tinker. It makes very interesting reading.

The latest research states that a longer line is not necessarily a problem

Parker’s research questions the conclusion that Paterson and Tinker drew from their exhaustive studies of how people read, carried out over 80 years ago. They were probably the first researchers to state that longer lines resulted in more corrective eye movements, thereby making reading slower:

Paterson and Tinker (1940) found that longer lines result in more regressions than shorter lines and these are “chiefly at the beginning of each line” (p576), i.e. after a return sweep. This was probably one of the earliest studies to report this finding. Later studies confirmed these results and clarified the nature of the regressions, describing them as corrective saccades

According to this argument, when the reader’s eye sweeps from the end of one line to the beginning of the next, it becomes more difficult to accurately locate the correct place in the text the further the eye has to travel. Readers “undershoot”, landing to the right of the left edge of the text (in left to right languages). As a consequence, they have to perform more corrective eye movements (known as regressions) to locate the start of the line, which makes reading slower.

Practically, this means that lines shouldn’t be too long. As to what we mean by too long, Tinker would propose a “safety zone” of around 45-90 charcters per line, which is what Butterick recommends in Practical Typography to this day.

Parker’s review of the research suggests that this is wrong. He doesn’t question the fact that longer lines result in more regressions, only that this does not necessarily mean that reading time is increased. Instead, when we undershoot, the eye still takes in information from the point where it lands, which means we can skip over this part of the line when we return to it. Overall, reading speed may actually be increased.

So should we all make our lines longer?

Possibly. After all, we’re doing our readers a service by making text quicker to read. Longer measures may also be handy from a design point of view, enabling us to use the space afforded by bigger, higher resolution screens, as well as larger font sizes (assuming this is a good thing in and of itself). Dyson ends on a call for us to experiment:

… it would be exciting if designers were to challenge received wisdom, explore the unfamiliar, and work on overcoming any possible reluctance from readers to engage with different ways of presenting text.

I would strike one note of caution. Most of this research is concerned with reading speed, rather than comfort. Obviously, speed can be an important aspect of reading, and may be related to comfort, but when I explored line length I was, dear reader, only interested in making your experience more pleasant by reducing eye strain and allowing you to sit back and relax. Dyson does cite research which concludes that readers at least perceive longer lines as more difficult:

One of the more consistent outcomes across studies is that long line lengths are least preferred or judged as least easy to read (Bernard et al., 2002; Dyson & Kipping, 1998; Youngman & Scharff, 1998). A moderate line length is preferred which would also be the most familiar format.

This would be a really interesting area of research. While reading speed is relatively easy to measure, it doesn’t reflect the full physical experience of the act, which is a more subtle and subjective thing. When we write we convey two messages – that of the text itself, and that created by the reading experience. In print, this signalled message is less pronounced as books and magazines look more or less the same, having settled on a standard set of professional type conventions decades ago. The fluid, untameable, open-to-anybody web, however, remains a wild typographical west, despite our best efforts to establish a set of rules.